Feb 27, Th 9am PT Build and Run Your First Eval

The TypeScript Agent Framework

From the team that brought you Gatsby: prototype and productionize AI features with a modern Javascript stack.

loved by builders, backed by founders

Just built a Web Search AI Agent using @mastra_ai Switched from Python to TypeScript for AI workflows 🚀@theananttrivedi
I've built a demo integrating RAG search using @mastra_ai with @vercel commerce and it's incredible.@iamrubenglez
More AI devtools need to be built natively for Typescript. check out @mastra_ai@davecyen
Exploring Web Search AI Agent using @mastra_ai right now... Working with Agents is boring without a proper layer like a framework.@alexisbchz
Plangrid
Plangrid
Plangrid
Plangrid

/agents

Build intelligent agents that execute tasks, access knowledge bases, and maintain
memory persistently within threads.

1
const chefAgent = new Agent({
2
  name: 'Chef Agent',
3
  instructions: 
4
    "You are Michel, a practical and experienced home chef" +
5
    "who helps people cook great meals."
6
  model: openai('gpt-4o-mini'),
7
  memory,
8
  workflow: { chefWorkflow }
9
});
Unified provider API

Switch between AI providers by changing a single line of code using the AI SDK

Memory

Combine long-term memory with recent messages for more robust agent recall

Prompt tuning

Bootstrap, iterate, and eval prompts in a local playground with LLM assistance.

Tool calling

Allow agents to call your functions, interact with other systems, and trigger real-world actions

/workflows

Durable graph-based state machines with built-in tracing, designed to execute complex
sequences of LLM operations.

1
testWorkflow
2
 .step(llm)
3
   .then(decider)
4
   .then(agentOne)
5
   .then(workflow)
6
 .after(decider)
7
   .then(agentTwo)
8
   .then(workflow)
9
  .commit();

.step()

llm

.then()

decider

.then()

agentOne

.then()

agentTwo

.after()

workflow

Clear control flow

Simple semantics for branching, chaining, merging, and conditional execution, built on XState.

.suspend()/.resume()

Pause execution at any step, persist state, and continue when triggered by a human-in-the-loop.

.watch() for real time state

Stream step completion events to users for visibility into long-running tasks.

Agent/workflow orchestration

Create flexible architectures: embed your agents in a workflow; pass workflows as tools to your agents.

*rag

Equip agents with the right context. Sync data from SaaS tools. Scrape the web.
Pipe it into a knowledge base and embed, query, and rerank.

.embed()

.query()

.rerank()

Unified vector store

Consistent API interface to upsert, index, and query data across providers

Metadata filtering

Narrow down your search space by querying on sources, time periods, or other properties

Agentic RAG

Equip agents with a vector query tool so they can search your knowledge base

*ops

Track inputs and outputs for every step of every workflow run. See each agent tool call
and decision. Measure context, output, and accuracy in evals, or write your own.

Performance metrics

Measure and track accuracy, relevance, token costs, latency, and other metrics.

Evals

Test agent and workflow outputs using rule-based and statistical evaluation methods.

Tracing

Agents emit OpenTelemetry traces for faster debugging and application performance monitoring.

a table showing a list of traces data on Mastra cloud